Watercolor [2026]: How to Start Without Wasting Money on Bad Supplies

By 2026, AI tools are embedded in almost every student's workflow whether schools have official policies about them or not. The debate about whether students should use AI has largely been superseded by the reality that they do. The more useful conversation is about how to use these tools in ways that genuinely help you learn and build skills versus ways that feel productive while actually hollowing out your own development. I've talked to enough students and instructors to have a real picture of both sides.

What AI Actually Does Well for Students

Explaining concepts is where AI tools genuinely shine for learning. If you're stuck on a concept from a lecture or textbook — a specific theorem, a historical event's causes, a chemistry mechanism — being able to ask for an explanation in plain language, ask follow-up questions, request analogies, and ask for the same concept explained differently until it clicks is enormously valuable. This is better than Googling in most cases because it's interactive. You can say "I understand the first part but not why X leads to Y" and get a targeted response. Khan Academy's integration of AI tutoring has shown real learning gains precisely because of this interactive explanation capability.

Brainstorming and outlining are genuinely useful applications. Using AI to generate a list of possible thesis angles, get feedback on whether an argument has obvious gaps, or build out an outline from your rough ideas is using AI as a thinking partner rather than a replacement for thinking. The critical skill here is that you bring the ideas and judgment; the AI helps you develop and stress-test them.

Editing and feedback on your own writing is valuable when used correctly. "Here is my paragraph — what is unclear, what logical gaps exist, and is the argument convincing?" is a legitimate use. "Write this paragraph for me" is not, for reasons that go beyond academic integrity.

Where AI Use Undermines Your Own Development

The core problem with using AI to generate work you submit as your own isn't just the academic integrity issue — though that's real, and AI detection tools have improved substantially. The deeper problem is that the struggle of producing work yourself is where the learning happens. Writing a first draft of an essay forces you to organize your thinking, identify what you don't actually understand, and develop an argument. Having AI write the draft and then lightly editing it produces something that looks like an essay but doesn't produce any of those cognitive outcomes in you. You can tell when this has happened because you can't explain or defend the content in a conversation about it.

The same applies to problem-solving. Using AI to get the answer to a math or programming problem, then copying it, means you've completed the assignment without building the ability to solve similar problems independently. When the exam arrives and the AI isn't available, the gap becomes painfully obvious.

A Framework That Actually Works

The test I'd apply to any AI use in studying: does this use require me to understand the material, or does it allow me to avoid understanding it? Asking AI to explain a concept until you understand it — then testing yourself by trying to explain it back without looking — requires understanding. Asking AI to summarize a chapter you haven't read and submitting notes based on the summary allows you to avoid understanding. The first builds the knowledge you need for exams and for actually knowing the subject. The second doesn't.

Specific tools worth knowing: Perplexity is excellent for research starting points (it cites sources, which you can then actually read). Quizlet's AI features for generating practice questions from your own notes are genuinely useful for active recall practice. Grammarly's revision feedback is legitimate editing assistance. Claude and ChatGPT for interactive explanation of concepts you're stuck on is among the highest-value uses.

From experience: Observing learning outcomes across different approaches and learners, the methods with the most consistent results are almost never the most novel — they are the ones that incorporate retrieval practice, spaced repetition, and genuine application.

Meta-analyses published in Psychological Science in the Public Interest found that retrieval practice (self-testing) produces approximately twice the long-term retention of re-reading — yet re-reading remains the most commonly used study technique among students at every level.

What Doesn't Work Despite Popularity

Re-reading highlighted notes — the most common study technique — is one of the least effective methods by research standards. It produces familiarity without producing durable memory. The discomfort of self-testing is precisely the signal that genuine learning is occurring, which is why students consistently underuse retrieval practice even when they know it works better. Feeling productive and being productive are different things in learning contexts.

Honest Bottom Line: Using AI as a tool for understanding is a powerful learning aid. Using AI to replace understanding is deceiving yourself. Test: can you explain the content without AI? If yes, you used it correctly. If not, you took a shortcut.